سامانه همایش‌ها TSTA | ثبت‌نام و ارسال مقاله کنفرانس‌ها و کنگره‌های ملی

صفحه اصلی / سی و دومین کنفرانس ملی و دهمین کنفرانس بین المللی مهندسی زیست پزشکی ایران

Super-Resolution Generative Adversarial Network for Photothermal Optical Coherence Tomography Signal Enhancement

نویسندگان :

Amirhossein Osooli¹ Mohammadhossein Salimi²

1- Tehran University 2- Tehran University

کلمات کلیدی :

Optical Coherence Tomography،Photothermal Imaging،biomedical imaging،generative adversarial networks،signal processing،super resolution

چکیده :

Photothermal Optical Coherence Tomography is an emerging functional imaging modality that combines the high spatial resolution of OCT with molecular specificity through photothermal contrast. However, PT-OCT signals typically require extensive temporal averaging to achieve adequate signal-to-noise ratios, limiting imaging speed and real-time applications. This work presents a novel one-dimensional Super-Resolution Generative Adversarial Network (1D-SRGAN) that reconstructs full-length PT-OCT temporal responses (176 data points) from truncated initial signal segments (22 data points), achieving an 8× temporal super-resolution while preserving critical spectral characteristics at the modulation frequency. The proposed architecture employs a deep residual generator with skip connections and adversarial training, validated through comprehensive frequency-domain analysis. Statistical evaluation demonstrates significant improvement in signal reconstruction fidelity, with preserved modulation frequency characteristics and enhanced signal-to-noise ratios compared to conventional interpolation methods.